Viktor
00:00:00.000
And especially if you're talking about Junior, that needs guidance from somebody, and I think that that's the key, right? You're junior, let's say software engineer, right? You enter the company, you're not making decisions. You are not. Changing the architecture, you're doing what you're told, That's very often the junior job and what I just described, if you replace the word junior with agent, we get the same result.
Darin
00:01:31.150
There are two types of developer job markets happening right now on one, which we've seen variations of before. You have companies like Anthropic paying their data scientists $750,000 or more just to work within the company. The other side of the market, you have Amazon laying off 16,000. You have block laying off either four or 6,000. I've lost the number. People are losing jobs. How can somebody try to stay employed in this type of market?
Viktor
00:02:11.026
I mean, so if every second person, loses job and I'm exaggerating now, then the one that. kept the job statistically is more likely to be the one that is better at the job. Nothing to do with ai.
Darin
00:02:26.406
right, this is the typical boom and bust cycle we can go back to again, trying to remember my years, the.com to dot bomb era of the 2000 early two thousands. You had a bunch of hiring in 99, 98, and all of a sudden, once the sock puppet went upside down, everybody got laid off.
Darin
00:02:53.638
It was temporary. So then we started going back up again because, oh, we found the real uses for whatever it was that we were building in 2000
Darin
00:03:03.829
Then, we had a variation of this again during. What? 2020? That's when it started happening. In March of 2020, people started hiring, hiring, hiring, because people were available because nobody was going to the office and people thought they had money until 2022 when they started firing. Everybody.
Darin
00:03:23.344
Just right sizing things. And again, in 2026, we're doing that cycle again, and I'm sure there were a number of cycles between 2020 20 that I just don't remember. I don't remember anything major for the tech space in that time. Financial services, absolutely. There was lots of things happening then, but. If people are trying to, it's like, okay, I just need to go back and get my master's so I can go get a job. Going to school to get a Master's isn't cheap either. How are you gonna pay for that? Because typically if you were gonna get a Master's, you would stay at your company. They would, as long as you stayed with them for at least five years, they would pay for your schooling because they would get extra value out of you. But come on. How many places are existing now for more than five or 10 years?
Viktor
00:04:07.550
The only thing I can say is say, Hey, if you go to school, you're doing the right thing because you know the low skilled jobs go out first. If AI is going to overtake all our jobs, then they will still need a few, few overloads or helpers, be that one.
Darin
00:04:27.557
Either be that one or become an electrician or an HVAC technician or a plumber because all the data centers need those people to build and maintain.
Viktor
00:04:44.508
So, uh, and I'm going to re exaggerate now. Ridiculous, but kind of, Hey, uh, your job could be whenever the, this drive breaks, you take it out and put the new one in Iraq. I think that that's, very, we are very close to have that. I mean, that's conceptually, and I know, I know it's not that simple, but that's conceptually not that different from Amazon Warehouse work that is getting heavily hammered right now.
Darin
00:05:15.577
Well, Amazon warehouse is robotics, so again, a robot could go and replace a hard drive. It's possible.
Viktor
00:05:25.309
exactly. I feel that the more, less creative jobs are going first, I think that what will be value distaste can you distinguish good idea from a bad idea? Can you come up with an idea? We are nowhere close to that being, ai, right? So, and there's much more, to it than that. But, the simpler the job is, the more likely it is that it'll go first.
Darin
00:05:56.562
Well, I've got a few stats around that. talking about entry level jobs, especially people maybe coming straight out of college or coming out of Boot camps. entry level tech positions are down 67% since 2022. Round numbers that feels about right.
Darin
00:06:12.187
Employment for developers age 22 to 25, down 20% from peak. But again, I'm gonna throw that out because peak was 2020 to 2022
Darin
00:06:23.707
that was just whatever. This one though, junior developer roles are dropping down by 40 to 50%
Viktor
00:06:31.915
Yes. That's a real problem and that that's pretty much in line what I said earlier, kind of the simpler the job, the more difficult to retain it.
Darin
00:06:43.203
in theory, AI is exceptionally good at implementation details, writing boilerplate, generating standard functions, debugging, syntax, all the stuff we would typically automate, we having AI do, which is what the juniors used to do. 'cause we would dump it off to them. So we could still do that. It'd be great, but by the time the junior gets it done to the level that we need it to be at, AI would've had it done three months earlier.
Viktor
00:07:09.060
Yeah. And especially if you're talking about Junior, that needs guidance from somebody, and I think that that's the key, right? You're junior, let's say software engineer, right? You enter the company, you're not making decisions. You are not. Changing the architecture, you're doing what you're told, That's very often the junior job and what I just described, if you replace the word junior with agent, we get the same result. I feel that people don't distinguish those things, but agents are really good at doing what we tell them to do today. When coding, when software engineering is concerned, they're pretty, pretty good. Now, when they mess up, when they don't have sufficient information, kind of like, Hey, build this feature. if you really detail it, the more you detail it and explain what you want and how it should be done and so on and forth, the better the job it'll do. Would you agree with that
Viktor
00:08:18.840
There we go. Bingo. And that's the problem we're facing. So really the question is how do you skip being junior? I can no answer to that one. Just to be transparent from the very start. But that's the question we should be asking kind of Better what Junior is is changing. Maybe Junior is the current senior.
Darin
00:08:43.065
A different way to ask that maybe is if AI is really great at doing the grunt work, how do we go about training our juniors now because the juniors were the grunts. I mean, used to, you could do apprenticeships. Wasn't that, there weren't that many informal, I should say not informal, but formal apprenticeship programs. There were a handful but unlike the traits where there are strong apprentice programs, just never really cared over software as much. I mean, sure, I could somebody send somebody to an apprentice program, but I'm not gonna send 'em to an apprentice program today. To do the things the way that we did them 10 years ago. That just doesn't make any sense.
Viktor
00:09:23.322
Here's the thing. I feel that in software engineering, uh, I'm not going to touch other professions. Just, uh, let's take with software engineering, the theoretical things you learn are the value is close to nothing. You start learning when you start doing stuff for real. I'm not sure where you stand on that one, but, I feel very strongly with it that, when you leave college, you know, close to nothing. You start getting real knowledge, experience, true experience by working on real projects. if it's true that. experience is gained by doing things, and everybody, including juniors, can do things much faster. Now with ai, that means that juniors are getting promoted from the knowledge perspective to something else much faster than before as well. If what makes you become a senior from junior is five years of experience, which would, which would be translated to, I dunno, how many lines of code and deployments and what's not. And you can do that in a year. Now that means that also the, timeframe you are getting promoted from the knowledge perspective is much shorter as well.
Darin
00:10:37.868
but then that's leading into, okay, you're getting promoted to a mid-level, but what is today's mid-level? They're too experienced. They're probably individual contributors, basically. Pure code living the ID all the time and bottom line, too expensive. So not only are we not bringing in as many juniors as we had, we can't justify the expense of a, maybe this is where we bring in the juniors. We can't justify the expense of a mid-level person because I could properly train a junior with AI and replace that mid-level easily.
Viktor
00:11:17.515
Maybe the problem is how we understand those roles because. Traditional company. Nobody will tell you using those words, but that's the meaning is what I'm about to say. And that's that you are a junior. Your job is not to think, your job is to do what you're told. And if that's the job, you're elite juniors, that that means that we need to somehow change that kind of like your job is to think from the very start.
Darin
00:11:50.991
But they were never taught to think. That might be the response. It's like I was taught to do algorithms not to think
Darin
00:12:03.508
is this. Yet another variation of shift left you're in a company now trying to work, but you got that job based on your knowledge that you learned while in school, and school isn't preparing you for work.
Viktor
00:12:17.311
if you think about it, it's actually just continues the trajectory that we've been on for decades now, and here's the trajectory and that's that. I mean it all started in a very different way, but eventually, we got to the idea that. You have many teams each team having people in plural and all those teams together eventually deliver something to production. it's not how the software industry started, but that's where it got at one point. And ever since we've been trying to reduce that. Agile was all about, okay, so can, can we actually put everything in one team, can this single team of relatively small number of people do development and testing and product management. We forgot about operations. We, we, we were not aware that those exist, but hey, there we go. And ever since it started, continues shrinking. A platform engineering is all about empowering a team, a single team, to do everything. That's what it's all about. So it key, it kept shrinking, and now we are getting to the, I'm sure that we are getting to a person, a person can do everything to bring this idea to life. And juniors have a role in that because they can have ideas. Sometimes they have better ideas than, than we do.
Darin
00:13:55.155
I'll correct you on that. Most of the time they probably have better ideas than we do nowadays.
Viktor
00:14:00.182
Yeah. I'm just, you know, hope never dies. I kind of, we, we are still somehow useful. So gimme that one.
Darin
00:14:06.270
It's interesting that you brought up platform engineering. The ideas of the next level of that's probably gonna be agent platforms almost goes unsaid.
Darin
00:14:18.537
I also wanna go back to, I was going to be snarky. It's like, well, microservices caused all the problems, but it's not that microservices caused the problems. It was because of the organization was modeled into what you said. They were team teams made up of multiple people therefore, the software MO was modeled after what the organization looked like. So it was natural to go to microservices because we wanted to have lots of separate teams. We wanted to have all the two pizza teams.
Viktor
00:14:46.851
Yeah, in full control of a service, and this is important, all the Ventil production. If you're not doing that, then microservice is not a good choice for you. I can tell you that in advance.
Darin
00:14:58.785
you've wasted your time and money and your organization. It, it's just, just be a monolith and be done if your organization is a monolith. Have monolith software, you'll be much happier. Maybe. Why does all this matter? I mean, obviously being employed and being able to pay bills. I saw a, uh, a cartoon the other day and I forgot the, it's working Chronicles, I think is what it's, and the question two people sitting in the frame, it's like, well, do you like your job? The answer was no. Well, why are you still there? I like paying my bills.
Darin
00:15:34.204
sometimes it doesn't just boil down to that. You just have a job to pay the bills. let's not, again, to company's your family, blah, blah, blah. 'cause that's all lies. But unless your company is all your family members, but then that gets really weird, really fast. what can we do to sort of like, we're being doom and gloom in all this for especially the, entry levels and the mids, the seniors, I have a feeling that they've actually gotta start doing work now instead of going to play golf every third day or every second day. Because if you can have a CC level person actually bringing a prototype that's like, Hey, I did this this weekend with my kid. Why is it taking us six months to do this other thing?
Viktor
00:16:15.272
As you mentioned seniors. I am not sure whether I would place Betson seniors or juniors if history can teach us something, the short one of this industry is that, that it's very likely the seniors will be rejecting all those changes that are happening right now. I know better. No AI can red code as I do. No AI can, SSH as well as I can, SSH into a machine. If that happens, actually our conversation should switch and that's what should we do with seniors.
Darin
00:16:51.924
Yeah, because if they're unwilling to make any kind of changes, they're effectively dead weight to the organization.
Viktor
00:16:59.156
And not just any drastically changes. 'cause effectively your job is completely different. You're not writing code anymore. And that's something you've been doing for 30 years, man.
Viktor
00:17:17.216
So I wouldn't be surprised that actually we end up, in many cases, situations where actually juniors say, oh, this cool. It's cool and a year later, kind of like I'm ninja and that senior that, you know, the one that can write code, prettiest code you ever saw in your life is still making it pretty. I wouldn't discard that as an option at all.
Darin
00:17:45.525
no, I think that's true. If they're not able to pivot to. Basically reading reviews of AI code becoming effectively a code reviewer, maybe helping with if there is some people there training, but you're ba effectively becoming a manager of X. Your value is no longer writing code. It's just not. And if you've been in an organization for 10 years and you don't understand the business, you need to get out anyway. 'cause if you don't get out, you'll be gotten out. Is a way to think about it.
Viktor
00:18:19.675
In the previous episode, I mentioned my experience in one of the companies where we were blocked by needing SH key, and that took a week or longer. And that's probably because there was a person in some team somewhere that does that. And let me tell you a secret about SH Keys. Anybody can create them. Anybody. Now, how likely that if that person could not transform from SSH key maker at that time to something, I'm going to say more productive. How likely is it that that person will say, yeah, ai. Cool.
Darin
00:19:11.805
in all this doom and gloom based on some numbers that I saw from Gartner. Companies that have platforms like true platforms in place, IDPs, all, all the different platforms in place and actually executing on them, they still need people because the rest of the company, that's what they're using. It takes me back to the early days of platform as a service. Not just infrastructure as a service, but platforms, a service that was life changing, basically. Heroku, right? And others. those were life changing things and it took humans to get there. It still takes humans to manage those. I'm gonna go against what you said earlier. I still think it'll take humans to manage those platforms because there's lots of, and going back to, again, back to the episode before. Where I need to bootstrap something. Well, I still need a human to bootstrap stuff, but the way things bootstrap from day to day may change because we upgraded from version X to version Y, which caused this cascade of something else that needed to be done.
Viktor
00:20:14.617
Small correction there. Sorry. We still need person probably to choose. The type of the bootstrap, they need to initiate the bootstrap
Darin
00:20:29.389
Oh, correct, yes. I misspoke that. Yeah, because there is some knowledge that needs to be applied there, but once that knowledge is figured out, it's like, okay, go off and do it again. It's just automation.
Viktor
00:20:42.014
How do you traditionally perform the, go ahead and do it. Hey Joe. You know the guy that just joined a week ago? Okay, do it. That's the job
Darin
00:20:56.772
Have you ever seen a shift like this happen this fast in your career? We've talked about this a little bit before, but
Viktor
00:21:02.236
this fast. Never. I've seen those shifts like this. I, I, people look at me like I'm crazy, but I'm saying that all those things happening now, were happening many times before with very similar effects. The differences in speed and scope. We've never seen this wide scope and we've never seen it happening so fast. But we've seen it happening before many times.
Darin
00:21:30.221
use two, it's rack and Stack servers. Then we got VMs. Slicing up the servers. And then we've got Kubernetes, which is a variation of VMs, containers, right containers, which gave us the ability to do Kubernetes.
Darin
00:21:45.146
That may, well, uh, I don't know that that's correct, but we'll say that it is. So these patterns have played out in the past. This is just a new pattern that I don't think we've ever, I'm hesitant to how I'm gonna say this, but I'm just gonna say it the way I thought it. We've never had. A tech come forward this way before that can actually replace a human. I'm not saying for everything, but a human couldn't replace a human. If I was working at a company and I left and they had to backfill for me. The new person didn't do what I did plus more day one. They maybe were able to five years later,
Viktor
00:22:20.789
Maybe that's a good analogy actually, what you just said. There are two types of employees. Some where you say, ah, you're leaving. Okay, I need five minutes to replace you. And some that are leaving and say this is a loss. Those are the people that are going to thrive in this new era.
Darin
00:22:43.645
Right, because if you can be easily replaced with ai, this goes back to the, you're going to be, I think.
Viktor
00:22:53.327
If what you do requires experience in knowledge that many, many other people have, and models are trained on what many, many people did in public. Those two are the same, If you have specific knowledge, specific experience, then you are being complimented with, with ai, you're not being replaced.
Darin
00:23:17.622
I've seen some interesting things around vibe coding. There's job boards. They're built up around vibe coding. the core jobs that people are looking for in that translate intense into prompts. Assess and refined AI outputs deliver working software, less code typing, more software architect and strategist. That's what we want.
Darin
00:23:41.125
now, I don't like the concept of vibe coding. I still like it a little more structured. Is spec driven development the answer for some things? Probably for everything? No. crossing between the two is important.
Viktor
00:23:54.867
I feel that all those terms are popping up like mushrooms and very confusing. I think it's easier to explain it simply. You are instructing agent system what to do. And you're correcting it when it does it wrong. That means that you have some qualities that are irreplaceable, at least right now. Or let me put it the other way around. If there is some, a task, a thing to do that you can do with the vanilla agent like CLO coat. with a single instruction, you will not have job for long. ' cause that's easy to replace. Hey, I want this feature, build it, and I'll just ship it when it's done. And there are cases like that and we don't need people for that.
Viktor
00:24:53.683
Yeah. Now we can get to the point, and I intentionally said vanilla agent, we can get to the point where, hey, this agent with those servers, with those, uh, with the, that this knowledge base, with those skills, with this and that and so on and so forth, right? The whole system. I set up one shot works that I do believe in. We are getting there. But not like Vanil, vanilla kind of installed cloud code that tell build this feature is done. We don't need that person.
Darin
00:25:28.156
what you're saying is at some point we'll be able to put in a one shot of, with, with all that infrastructure built around it, infrastructure being code skills, mcps, all the things
Viktor
00:25:39.884
All the knowledge. All the knowledge, all the everything in my head, everything the company has right. And it's important that it's not only what company has, but my head, kind of like my way of thinking. If my way of thinking and operating produces better results than your way of thinking and operating, then I'm ahead of you then. I don't need you. You need me, right? Hypothetical, you hypothetical me.
Darin
00:26:08.999
And the hypothetical to continue on is eventually we're gonna get to the point to where. Hey, uh, Bob. We'll call our AI Bob, say, Hey Bob, we've got a new competitor over here. Take their website, clone it, add these six new features. I'll see you in the morning. Have a good night, and you come in the next morning and it's there and running and already signing up people. That's possible. Probable, I don't know, but it is possible.
Viktor
00:26:40.545
So it all boils down who is better at figuring out what to do and how to do it and when to do it. That's the job. And if you think about it, what makes one company more successful than the other? And very often the answer is not because they have more money. Because if that would be the case, we would never have startups, So we have startups that many of them fail, but many of them become new. Massive big companies right from ground up and what differentiates them from whomever they're competing with, the big guys is that they have a better idea, better way to execute that idea, better way to plan that idea, and so on and so forth. That's the value. So think of it like a startup, right? What do you have? You, you don't have the, the manpower and money that, companies you're competing with have.
Viktor
00:27:37.914
You need to change the game. You need to think differently think differently is not something I expect from AI anytime soon.
Darin
00:27:44.773
Speaking of soon, I'm gonna take us back. Do you think the term or the the title vibe, coder. Is gonna be any different than the term one? I haven't heard in years. I was thinking about webmaster. once the whole industry sort of catches up, that's just gonna sort of fall off and it'll be okay. It's sort of like when we had, what was it a couple years ago? prompt engineer. That was the phrase I was just thinking about. It's like, pro, what the blank is that, you know, it's like now it's not a thing anymore.
Viktor
00:28:13.987
I would rather use context in context engineer or context master, or whatever you wanna call it. And while it's not a great term, I think it, it explains things much better than, vibe coding or. Prompt engineering, because prompt is one thing that you send to the agent and that gets combined with tools and, uh, from context, from tools, from context, from system context and so on and so forth, right? it's a one tiny piece in a very big picture. but ultimately all the countries combined one way or another. Through input, through static from databases, from tools and so on and so forth, is what makes LM do what it does. So some kind of wording that says that, hey, um, it's primarily about context. And context can be many things just to be a hundred percent cloud transparent, like storing, getting data into a database and fetching it from there so that you populate part of the context. Again, running a script that gets the output, that will go into the context as well, So it is really about generating that context that will allow LLM to say, this is the solution. That's the real work.
Darin
00:29:51.294
Yeah. Or ai, augmented developer. Take your pick. Either one works. It's like, so somebody, they're just popping into cloud code. They're doing their thing every day. What? What is their, what does that person's day look like?
Viktor
00:30:04.831
that feels like what I'm doing most of my time these days, and that's, Hey, what should I work on today? pick a job start talking with your favorite friend. In my case, Claude.
Darin
00:30:20.062
So a, a. Highly functioning ai, augmented developer is treating it like pair programming.
Viktor
00:30:28.801
Yes, that's my case. I, I never reached the point of one shot something. Never. I have a criteria even how I will say that I reach that part when I have a session that's. From beginning to the end of something meaningful where I said yes to everything. that's the day when I will turn, to do one shot things, at least for something. I never reached that part. I'm not saying that the others haven't. I'm not saying it could be on me just to be a hundred percent clear. Maybe I'm not doing it right. I never had a session where I haven't said a single time, no.
Viktor
00:31:10.949
Excluding something ridiculous kind of, Hey, uh, here's a security report from the CI pipeline that just executed when I created a pr fix it and goes there and does something. And I say, yeah, kind of. Yeah, it's fine. I mean, I'm, I'm now talking about more meaningful work.
Darin
00:31:28.863
the progression I'm hearing is you might be a, a one shot attempt that then turns into a pair programmer, but then eventually turns into a manager of a team of agents. That's, that's sort of the natural progression that I think everybody needs to go through to understand, because each one builds on the other.
Viktor
00:31:51.254
Yeah. Pretty essentially the same as. manager before ai, right? What, what do managers do, right? They have two primary, three primary purposes, right? Figuring out what should be done and how should be done and when should be done and all this stuff. And then, uh, the team starts doing it and they look at the team as a whole. That would be second. And then they, work with individuals on individual basis as well. That's my very uneducated guess of what managers do, and that feels very similar to, to what I'm doing with, with agents right now. I want this go and do it like the, I mean, very long description. And then eventually I graduated and kind, now I have four of them doing different things and I come back every once in a while. They talk to each other because very often it's connected things. So they work as a team somehow? sometimes directly, sometimes through me. 'cause oh, you should maybe ask the agent in this project to do this because you're a frontend and that's the backend and all that stuff. I see it as my dream coming true. This is what I always dreamt. I always liked the idea of me being a manager. couple of times that I tried to do it, it was a complete failure and identified that people were the problem, not me. now I can go back to try the same thing without people and I love it, Claude never told me. Uh, no,
Viktor
00:33:40.520
It's extremely dangerous. Extremely dangerous, and that's where experience we were talking about from before actually comes into play, It's very easy to be misguided by models, not necessarily by them suggesting or doing the wrong thing, but simply, that thing does not really fit into what you're trying to do. And just like teams are very, most of the time teams do not have a hundred percent grasp of what is in the head of a manager, right? Uh, because it's close, impossible for that manager to explain every single detail that is in their head. Same thing here. Then distinguishing good from bad, right from the wrong path. that's most of my job right now.
Darin
00:34:24.291
Well, those are the skills that will keep you employable, I think, is understanding. How do I work with, fill in the blank people, agents, vendors, clients, being able to do those things and adapt is gonna keep you employable if you choose to not adapt. The chances of your employment are tenuous at best.
Viktor
00:34:51.808
Look, I'm, I'm very close to, and people communicating with me will start noticing it, not so, far away. In the future, I will be delegating more and more of my communication to agents just to be completely transparent, with my guidance. And if that happens, I can only expect that actually others will do eventually the same. And that means that I will not be talking to vendors in the future. I'm using vendors as one of the examples you mentioned.
Viktor
00:35:33.743
that's what's coming. And I know the people, I think, will think that this are going too far, but I, I don't think that we are even far from there. Okay. Sorry. Not, not from there. The world doesn't change that fast. But from seeing successful implementations of that on smaller scale, uh, we are getting there very, very fast. How long will it take until, You respond only to a fraction of emails you are currently responding yourself.
Darin
00:36:02.725
That one's hard for me. I'm not willing to give that one up yet. I'm getting close.
Viktor
00:36:09.524
I, I'm not giving up, I'm not doing that today. I'm just saying that I feel that we are very close to that at the time, not we, I am very close to. And this is a ridiculous example, but delegating response to a fresh, some part of my email. So something that is, repeat. I, I get a lot of repeat questions like, here's an example. I'm pretty sure that there is a number of emails you receive somehow related to this podcast that you copy and paste the answer.
Viktor
00:36:43.824
how big of a stretch is it from you copying and pasting the answer to instructing agent. Okay. When, when it's this type of query, this the response, it's not even complicated. One, it.
Viktor
00:37:00.524
I am not necessarily saying today, but I'm being kind of pessimistic now. It is getting very close, right? Some people are probably doing it already with DOPP and Claw and what's not, but I'm not. You're probably not. But, I can imagine finishing this year like that.
Darin
00:37:13.739
It's very possible talking about some skills. If you can do systems thinking, that's a good thing. If you understand business domains, you'll probably stay. Uh, this, again, this goes back to the senior concept I had earlier. If you've been at a company for 10 years. And you don't understand what the business does. You don't need to be there anymore. Sorry, you don't,
Darin
00:37:37.919
and there's a handful of other things that, but these are again, the things that we're, we're talking about here. Look at my notes. System design, orchestration, security and governance, domain expertise, communication and judgment. If you know how to do that within an organization, within the rules of the organization, the chances of you getting. Cut. I'm not gonna say zero, but they're probably minimized a bit, meaning it went from zero to 10%. How can we help people understand, okay, you've spent the past 30 years learning how to use Eclipse or IntelliJ. Uh, you need to change everything now.
Viktor
00:38:19.870
Let me phrase it this way, and this applies only with people with more than a decade of experience. If you're a type of person that changed what you do and how you do it multiple times in your career, you're fine.
Darin
00:38:36.290
In other words, if you've been the person that's been moving jobs every three to five years, you're probably okay.
Viktor
00:38:41.139
Doesn't have to be moving jobs as if different companies, it can be the same company, but kind of, Hey, if you started with as DB two expert administrator and you're still D to administrator, you, I'm surprised that you still have a job.
Viktor
00:39:02.388
Like to, to get with mainframe bodies. but you get the point, right? Uh, there are people who are changing, right? Kind of that embrace, embrace cloud, embrace the containers, embrace Kubernetes, and so on and so forth. And there are still people fighting it. people who, made those changes multiple times before. I'm not sure why this would be any different. so here's the thing, and this is where I think maybe age or experience might come in play. I often get, questions or queries in relation to ai, and I feel behind those Cs. Ah, you're saying ai, uh, no, no, no. Kubernetes me or ai I, no, no, no. Go me. And I feel that very often behind those queries, that you are young. That's what you started with, and you think that that's the best thing, think ever. And it'll stay the same because you haven't experienced those changes before. You cannot relate to understand that when you Kubernetes, you don't know that we were saying the same 14 four mass before that, right? I was swearing as much for Mezo as now for Kubernetes, I, I have full rights to expect that something is coming around the corner again. But if your whole career is one thing, hopefully short career, if it's long than you're completely messed up. Uh, but if, if your whole career is one thing, then I understand how you cannot let it go.
Darin
00:40:39.356
Let's flip it around. Things you don't want to do, don't panic. Learn every AI tool that's out there because especially if you follow cloud code, you know there's at least one release a day every day of the week.
Darin
00:40:53.755
There's, there's new things every day, new fixes, new features, new everything. Don't become a prompt engineer as your identity, because that doesn't make any sense because we've already talked about that and it's like, that's just table stakes. You need to understand how to do context engineering. Biggest thing though is don't abandon fundamentals.
Viktor
00:41:18.276
Hmm, because you understand fundamentals, you will gain revelation. And that revelation being that actually all those new things coming to different tools in the, in a specific space, they're very, very similar. They're almost the same, So once you grasp how it all works, competition becomes very easy to understand.
Darin
00:41:45.985
If you were hiring somebody today for your team, if you had a team, a human, let's put it that way, you're hiring, trying to hire somebody and you're interviewing, what's the one skill you would want to find today that you wouldn't have asked about two years ago?
Viktor
00:42:02.352
because. A few times that when I was hiring and also when I was not hiring, but I had opinions who we should hire. I always valued capacity to learn and to adopt, then actual experience. I think that the right person can learn anything. And it's much harder to, it's easier to teach something specific to a person, then teach them how to be playful, curious how to learn. I had the story, I probably repeated it in this podcast, so Dar cut it. one time when I was hiring, it was for a relatively young role, junior role, and I did some kind of a test. I don't remember what it was, you know, build this like this and whatever. And they were doing it on my laptop, on a company laptop. And they had open, I think at that time was Eclipse and also open browser. The one who went, uh, that switched the browser and Googled got the job.
Darin
00:43:14.737
Because they realized it didn't need to be. From memory, this wasn't a fifth grade
Viktor
00:43:21.506
to the problem you gave me. That's what I'm looking for. I'm not looking for a person who has the solution to my problem because my problems are different every day, I want people. With the ability to solve problems, I don't care how, if you know the solution, great. But they were all young, so none of them had solution. None of them knew how to write code, while implemented TDD and to build it. It was a full, I don't remember the details, but I know it was something very simple. But the full shenanigans, kind of like up to imaginary production, none of them knew how to do that. And only one did not try to impress me how they know what they don't know. So that's, that's the person. I think there's the same person today, if not even more.
Darin
00:44:12.541
I think that reinforces what I'm getting ready to say. The job market's not coming back to where we were in 2021. In fact, that was so super inflated, it shouldn't have existed in the first place. A developer that is able to adapt and understand how to use AI as a force multiplier is gonna be the one that's able to pull things off. the one that's unwilling to adapt. Thinking, well, no. My free range organic code is gonna be better than anything ever created. Well, it might be better. I won't argue that point, but your free range organic code is not gonna be there faster than the code that I just made a million dollars with.
Viktor
00:44:52.000
Yeah. you know it's, not about code. It's about the features that that code enables and the features that are there and the features that are missing, and, uh, issues that are not fixed. And you, if you can do all those. And make it pretty or whatever quality, whatever your handcrafting is. Hey, I think it's great. It's just that I never saw it. Have you ever seen a project where there are no open issues older than a week or a month?
Viktor
00:45:29.902
Have you ever seen a where actually all the features that somebody wanted to have in that project were implemented?
Viktor
00:45:39.532
Now, is that better than maybe not so artisanal code, but actually where all those features are actually there?
Darin
00:45:49.793
No, we're trying to not make you feel so bad. We're just two guys on a microphone. But here's something you might wanna think about doing this week. Take an honest evaluation of where you stand right now. What's the chances of you getting laid off? Because AI could replicate 90% of what you do. have that, that meeting with yourself to figure out, okay, what's going on? Then maybe within the month, if you haven't started doing any kind of AI coding yet, crank up cloud code, crank up whatever, you know, spend 20 bucks for one month and try to build something non-trivial and see how it works.
Viktor
00:46:28.041
His additional advice, do not think that the results you'll get from it first week are the results you will get from it a year later. It's also about experience with those tools and systems and WhatsApp. And building what you're needing. You need to build the capabilities that you need. It's not all out the box. And to build a muscle memory and knowledge, if you think that this is not good enough, after a week, the the hands, I will stop. That's when you're making mistake.
Viktor
00:47:05.325
No, even better with Java. Your first week with Java. The, that result in a production scalable code that can be running and serving 1 million concurrent users, which is probably criteria that you have now, 10 years later.
Darin
00:47:21.924
And now you don't even think about it. It's like, oh yeah, we need to do this, this, this, this, this. This goes back to the, you know, systems thinking architecture. You just, you know what it is to the juniors out there or the new entry level. Don't give up. There will be companies that will hire you. Who they are right now. I can't tell you. I don't know.
Viktor
00:47:42.188
dear junior. There is a strong chance that you will replace of old farts like we are who are still doubting it. We are fine.
Darin
00:47:52.754
you'll be fine. Just don't give up. So what do you think? Do we stand any kind of chance to stay employed or not? No. I'm already retired, so no, I'm not employable to begin with.
Viktor
00:48:03.449
You stand zero chance to be employed. You don't even try to be employed anymore.
Darin
00:48:07.480
yeah, so head over to the Slack workspace, go to the podcast channel, look for episode number three five one and leave your comments there.